A regret-free (sublinear-regret) episodic algorithm for LTL objectives, built on optimistic interval-MDP value iteration for reach-avoid and a graph-learning preprocess requiring a known minimum transition probability.
Taming the monster: A fast and simple algorithm for contextual bandits
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Regret-Free Reinforcement Learning for LTL Specifications
A regret-free (sublinear-regret) episodic algorithm for LTL objectives, built on optimistic interval-MDP value iteration for reach-avoid and a graph-learning preprocess requiring a known minimum transition probability.